* Removed sandbox risk categories; feedback indicates that these are not
that useful and "less is more"
* Tweaked the assessment prompt to generate terser answers
* Fixed bug in orchestrator that prevents this feature from being
exposed in the extension
This fixes an issue where messages sent during the final response stream
would seem to disappear, because the "queued messages" UI wasn't shown
during streaming.
fixing drag/drop photos bug in codex
state of the world before:
sometimes, when you drag screenshots into codex, the image does not
properly render into context. instead, the file name is shown in
quotation marks.
https://github.com/user-attachments/assets/3c0e540a-505c-4ec0-b634-e9add6a73119
the screenshot is not actually included in agent context. the agent
needs to manually call the view_image tool to see the screenshot. this
can be unreliable especially if the image is part of a longer prompt and
is dependent on the agent going out of its way to view the image.
state of the world after:
https://github.com/user-attachments/assets/5f2b7bf7-8a3f-4708-85f3-d68a017bfd97
now, images will always be directly embedded into chat context
## Technical Details
- MacOS sends screenshot paths with a narrow no‑break space right before
the “AM/PM” suffix, which used to trigger our non‑ASCII fallback in the
paste burst detector.
- That fallback flushed the partially buffered paste immediately, so the
path arrived in two separate `handle_paste` calls (quoted prefix +
`PM.png'`). The split string could not be normalized to a real path, so
we showed the quoted filename instead of embedding the image.
- We now append non‑ASCII characters into the burst buffer when a burst
is already active. Finder’s payload stays intact, the path normalizes,
and the image attaches automatically.
- When no burst is active (e.g. during IME typing), non‑ASCII characters
still bypass the buffer so text entry remains responsive.
This PR adds support for a model-based summary and risk assessment for
commands that violate the sandbox policy and require user approval. This
aids the user in evaluating whether the command should be approved.
The feature works by taking a failed command and passing it back to the
model and asking it to summarize the command, give it a risk level (low,
medium, high) and a risk category (e.g. "data deletion" or "data
exfiltration"). It uses a new conversation thread so the context in the
existing thread doesn't influence the answer. If the call to the model
fails or takes longer than 5 seconds, it falls back to the current
behavior.
For now, this is an experimental feature and is gated by a config key
`experimental_sandbox_command_assessment`.
Here is a screen shot of the approval prompt showing the risk assessment
and summary.
<img width="723" height="282" alt="image"
src="https://github.com/user-attachments/assets/4597dd7c-d5a0-4e9f-9d13-414bd082fd6b"
/>
- ensure paste burst flush preserves ASCII characters before IME commits
- add regression test covering digit followed by Japanese text
submission
Fixesopenai/codex#4356
Co-authored-by: Josh McKinney <joshka@openai.com>
Extends shell wrapper stripping in TUI to handle `zsh -lc` in addition
to `bash -lc`.
Currently, Linux users (and macOS users with zsh profiles) see cluttered
command headers like `• Ran zsh -lc "echo hello"` instead of `• Ran echo
hello`. This happens because `codex-rs/tui/src/exec_command.rs` only
checks for literal `"bash"`, ignoring `zsh` and absolute paths like
`/usr/bin/zsh`.
**Changes:**
- Added `is_login_shell_with_lc` helper that extracts shell basename and
matches against `bash` or `zsh`
- Updated pattern matching to use the helper instead of hardcoded check
- Added test coverage for zsh and absolute paths (`/usr/bin/zsh`,
`/bin/bash`)
**Testing:**
```bash
cd codex-rs
cargo test strip_bash_lc_and_escape -p codex-tui
```
All 4 test cases pass (bash, zsh, and absolute paths for both).
Closes#4201
Expand the custom prompts documentation and link it from other guides. Show saved prompt metadata in the slash-command popup, with tests covering description fallbacks.
## Summary
- add a kill buffer to the text area and wire Ctrl+Y to yank it
- capture text from Ctrl+W, Ctrl+U, and Ctrl+K operations into the kill
buffer
- add regression coverage ensuring the last kill can be yanked back
Fixes#5017
------
https://chatgpt.com/codex/tasks/task_i_68e95bf06c48832cbf3d2ba8fa2035d2
keep a 1 cell margin at the right edge of the screen in the composer
(and in the user message in history).
this lets us print clear-to-EOL 1 char before the end of the line in
history, so that resizing the terminal will keep the background color
(at least in iterm/terminal.app). it also stops the cursor in the
textarea from floating off the right edge.
---------
Co-authored-by: joshka-oai <joshka@openai.com>
Refactor trust_directory to use ColumnRenderable & friends, thus
correcting wrapping behavior at small widths. Also introduce
RowRenderable with fixed-width rows.
- fixed wrapping in trust_directory
- changed selector cursor to match other list item selections
- allow y/n to work as well as 1/2
- fixed key_hint to be standard
before:
<img width="661" height="550" alt="Screenshot 2025-10-09 at 9 50 36 AM"
src="https://github.com/user-attachments/assets/e01627aa-bee4-4e25-8eca-5575c43f05bf"
/>
after:
<img width="661" height="550" alt="Screenshot 2025-10-09 at 9 51 31 AM"
src="https://github.com/user-attachments/assets/cb816cbd-7609-4c83-b62f-b4dba392d79a"
/>
Clear the history cursor before checking for duplicate submissions so
sending the same message twice exits history mode. This prevents Up/Down
from staying stuck in history browsing after duplicate sends.
Before when you would enter `/di`, hit tab on `/diff`, and then hit
enter, it would execute `/diff`. But now it's just sending it as a text.
This fixes the issue.
- prefix command approval reasons with "Reason:"
- show keyboard shortcuts for some ListSelectionItems
- remove "description" lines for approval options, and make the labels
more verbose
- add a spacer line in diff display after the path
and some other minor refactors that go along with the above.
<img width="859" height="508" alt="Screenshot 2025-10-02 at 1 24 50 PM"
src="https://github.com/user-attachments/assets/4fa7ecaf-3d3a-406a-bb4d-23e30ce3e5cf"
/>
## Summary
- show the remaining context window percentage in `/status` alongside
existing token usage details
- replace the composer shortcut prompt with the context window
percentage (or an unavailable message) while a task is running
- update TUI snapshots to reflect the new context window line
## Testing
- cargo test -p codex-tui
------
https://chatgpt.com/codex/tasks/task_i_68dc6e7397ac8321909d7daff25a396c
Here's the logic:
1. If text is empty and selector is open:
- Enter on a prompt without args should autosubmit the prompt
- Enter on a prompt with numeric args should add `/prompts:name ` to the
text input
- Enter on a prompt with named args should add `/prompts:name ARG1=""
ARG2=""` to the text input
2. If text is not empty but no args are passed:
- For prompts with numeric args -> we allow it to submit (params are
optional)
- For prompts with named args -> we throw an error (all params should
have values)
<img width="454" height="246" alt="Screenshot 2025-09-23 at 2 23 21 PM"
src="https://github.com/user-attachments/assets/fd180a1b-7d17-42ec-b231-8da48828b811"
/>
Fixing the "? for shortcuts"
- Only show the hint when composer is empty
- Don't reset footer on new task updates
- Reorder the elements
- Align the "?" and "/" with overlay on and off
Based on #4364
